AI Process Optimisation
Leading AI programmes — from the executive level.
AI process optimisation is not an IT task but a leadership task: business case, governance, risk, change. I lead your AI programme at C-level — and deliver it with your teams.
Many companies have AI pilots. Few have AI in the process. The difference rarely lies in the technology but in the question of who owns the programme: who decides which processes come first, who carries the risk, who makes sure the business units come along? That is the role I take — for a defined period, with a mandate.
Why AI programmes need leadership
The typical reasons AI initiatives get stuck at pilot stage have nothing to do with models:
- There are ten ideas and no business case that justifies any one of them.
- Nobody on the management team owns the programme — it sits "with IT" or "with the innovation team".
- Data protection, liability and traceability are addressed only once the first result already exists.
- The people whose work will change were never asked — and put on the brakes.
- Vendors sell platforms before the process is understood.
An AI programme therefore needs the same leadership as any other transformation: goal, priorities, budget, risk owner, reporting.
What I take on as interim AI programme lead
- Prioritisation: which processes have enough volume, error cost and data quality for AI to pay off — and which do not.
- Business case: benefit in error rates, rework, cycle time and operational risk — measurable before and after rollout.
- Governance and risk: roles, approvals, controls, handling of personal data, limits of automation.
- Vendors and architecture: choosing models, platforms and partners by requirements, not by demo.
- Change: involving business units, redefining roles, taking concerns seriously — so the solution is actually used.
- Reporting: progress, cost and risk in a form that management and the supervisory board can judge.
Example: AI agents in finance workflows
Since October 2025 I have been leading three parallel projects at an Austrian mobile operator in which AI agents are being built into financial processes. The goal is end-to-end automation that lowers error rates, manual rework and operational risk — and with them the customer queries and complaints that flawed processes generate.
My role is that of programme owner: setting priorities, bringing business units and technology together, resolving risk questions, reporting results to management. The work happens in the process, not in a lab.
Data protection, EU hosting, traceability
In finance and customer processes, AI can only be used if three questions are answered up front: where is the data processed, who is liable for the decision, and can an auditor trace what happened? My principles:
- Personal data is processed only in compliance with GDPR — preferably hosted in the EU and without retention by the model provider.
- Automation with defined limits: what does the system decide alone, what does a person confirm?
- Every automated decision is logged and explainable.
- Clear accountability for operation, monitoring and switch-off.
For supervised companies such as banks, the rules of outsourcing governance apply in addition — more under outsourcing and provider management.
From programme to delivery
I lead the programme, decide and take responsibility — but I do not build models single-handedly. Delivery happens with your teams and, where needed, with specialised partners. For AI consulting, software development and GDPR-compliant AI operations, my sister brand thinkai.at is available; C-level programme accountability stays here.
What you have at the end: productive processes instead of pilots, governance that withstands audits, and a team that owns the solution. All the executive roles I take on for a defined period are listed under interim management.
Over many years, Robert Pabeschitz has impressed me again and again in our joint projects through his drive for innovation and his customer focus.
FAQ
Frequently asked questions
Where do you start with AI process optimisation?
With the processes that cause the most manual rework, errors or waiting time today — and for which structured data exists. Finance processes such as invoice verification, reconciliations and complaint handling are often a good start because benefit and risk are measurable.
How long until the first productive result?
A cleanly scoped process can often be brought into controlled productive use within a few weeks. What takes longer is data access, approvals and involving the business units — which is why I start on those from day one.
Is AI in finance processes possible in compliance with GDPR?
Yes, if data processing, hosting and access are designed that way from the start. What matters is the choice of provider, the place of processing, the contractual basis and the question of which decisions a person confirms. I involve data protection and legal before the first line is built.
What does AI programme leadership by an interim manager cost?
Like every mandate, on a day-rate basis — sized by responsibility, scope and term. Programme leadership is usually a fraction of the overall budget and the part that decides whether the budget achieves anything. I quote terms after the intro call; see process and terms.
Do I need an interim manager or an AI consultant?
A consultant assesses options and recommends. An interim manager takes over the programme with budget, decision authority and a reporting duty, and brings it into operation. If you want to know what is possible, advice is enough; if you want it to happen, you need leadership.
Contact
Let's talk about your AI programme.
A first idea or a pilot that is stuck — in the intro call we work out which processes are worth it and what leadership the programme needs.
Confidential and non-binding.